Data processing system for determining similar users

Abstract

The invention relates to a data processing system for determining similar users, which comprises a processor and a memory in which a computer program is stored, and when the computer program is executed by the processor, the following steps are realized: obtaining an initial feature value list set corresponding to an initial user ID list, obtaining an initial feature vector list set corresponding to the initial user ID list, and obtaining a user ID list set corresponding to the initial feature vector list set; obtaining a candidate feature vector list set corresponding to the initial user ID list, obtaining a specified feature vector list set corresponding to the initial user ID list, obtaining a target similarity list corresponding to the initial user ID list according to the specified feature vector list set, determining similar users of the target user from the initial user ID list, and sending the similar users to the target user. According to the method, meaningless data are deleted in the process of obtaining the similar users, meanwhile, data with small influence on the similarity is deleted, the calculation amount of follow-up similarity calculation is reduced, the efficiency of obtaining the similar users is improved, and the accuracy of obtaining the similar users is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a data processing system for determining similar users. Background Art

[0002] As artificial intelligence technology gradually matures, its application scenarios become more and more abundant. At present, with the continuous development of medical informatization, the integration of artificial intelligence technology and the medical field continues to deepen. There are a large amount of data in the medical information service platform. These data are not only huge in quantity, but also diverse and complex. How to associate these data has become a hot research direction. For example: how to quickly and accurately find similar patient users is of great significance in user classification scenarios in the medical field. By obtaining similar patient users, accurate classification research reference data can be efficiently provided to relevant workers.

[0003] Common methods for obtaining similar users are: the first method is to first build a full data index of the user, then set the user's key features, and make judgments based on the keywords in the key features to obtain similar users. The accuracy of similar users obtained only through the keyword method is low. The second method is a similarity-based method, which first defines a similarity measure, and then screens by calculating the similarity between two users. This increases the time cost, reduces the efficiency of obtaining similar users, and makes it difficult to determine similar users in large-scale data sets. Summary of the invention

[0004] In view of the above technical problems, the technical solution adopted by the present invention is: a data processing system for determining similar users, the system comprising: a processor and a memory storing a computer program, when the computer program is executed by the processor, the following steps are implemented:

[0005] S100, obtaining an initial feature value list set A corresponding to the initial user ID list = {A1, ..., A i , ..., A n}, A i ={A i1 , ..., A ij , ..., A im}, where A ij is the jth initial eigenvalue in the initial eigenvalue list corresponding to the ith initial user ID, i=1...n, n is the number of initial user IDs, j=1...m, m is the number of initial eigenvalues ​​in the initial eigenvalue list.

[0006] S200, according to A, obtain an initial feature vector list set C corresponding to the initial user ID list = {C1, ..., C i , ..., C n}, Ci ={C i1 , ..., C ij , ..., C im}, C ij is the jth initial feature vector in the initial feature vector list corresponding to the i-th initial user ID, where C ij =(i, j, A ij ).

[0007] S300, according to C, obtain a candidate feature vector list set D corresponding to the initial user ID list = {D1, ..., D i , ..., D n}, D i ={D i1 , ..., D ir , ..., D is(i)}, D i is the candidate feature vector list corresponding to the i-th initial user ID, D ir D i The rth candidate feature vector in D, r = 1...s(i), s(i) is i The number of candidate feature vectors in , where A ij When C ij From C i to form D i .

[0008] S400, according to D, obtain a specified feature vector list set L corresponding to the initial user ID list = {L1, ..., L i , ..., L n}, L i is a list of designated feature vectors corresponding to the i-th initial user ID, wherein L is obtained in S400 by the following steps: i :

[0009] S401, according to A, obtaining a candidate feature identifier list G, wherein the candidate feature identifier list includes a plurality of candidate feature identifiers.

[0010] S403, when D ir When the initial feature identifier corresponding to the initial feature value in G is consistent with the candidate feature identifier in G, D ir Insert to L i To obtain L i .

[0011] S500, according to L, obtain the target similarity list N={N1,...,N i , ..., N n},N iis the target similarity corresponding to the i-th initial user ID.

[0012] S600, when N i ≥N 0 , the initial user corresponding to the i-th initial user ID is determined as a similar user of the target user, where N 0 is the preset similarity threshold.

[0013] Compared with the prior art, the present invention has obvious beneficial effects. By means of the above technical solution, the data processing system for determining similar users provided by the present invention can achieve considerable technical advancement and practicality, and has wide industrial utilization value, and has at least the following beneficial effects:

[0014] A data processing system for determining similar users comprises a processor and a memory storing a computer program. When the computer program is executed by the processor, the following steps are implemented: obtaining an initial feature value list set corresponding to an initial user ID list, obtaining an initial feature vector list set corresponding to the initial user ID list according to the initial feature value list set, obtaining a candidate feature vector list set corresponding to the initial user ID list according to the initial feature vector list set, obtaining a designated feature vector list set corresponding to the initial user ID list according to the candidate feature vector list set, obtaining a target similarity list corresponding to the initial user ID list according to the designated feature vector list set, and determining similar users of a target user from the initial user ID list according to the target similarity list. In the process of obtaining similar users, this embodiment deletes meaningless data and considers deleting data with little influence on similarity, which can reduce the amount of calculation for subsequent similarity calculation and reduce time cost, thereby improving the efficiency of obtaining similar users, preprocessing the obtained original data, obtaining feature vectors corresponding to users by using a hash model, and performing similarity judgment based on the obtained feature vectors, thereby improving the accuracy of obtaining similar users.

[0015] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the following specifically cites a preferred embodiment and describes it in detail with the accompanying drawings as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flowchart of a computer program executed by a data processing system for determining similar users provided by an embodiment of the present invention; DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0018] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products, or devices.

[0019] Example

[0020] This embodiment provides a data processing system for determining similar users, the system comprising: a processor and a memory storing a computer program, when the computer program is executed by the processor, the following steps are implemented, such as: Figure 1 As shown:

[0021] S100, obtaining an initial feature value list set A corresponding to the initial user ID list = {A1, ..., A i , ..., A n}, A i ={A i1 , ..., A ij , ..., A im}, where A ij is the jth initial eigenvalue in the initial eigenvalue list corresponding to the ith initial user ID, i=1...n, n is the number of initial user IDs, j=1...m, m is the number of initial eigenvalues ​​in the initial eigenvalue list.

[0022] Specifically, the initial user ID list includes several initial user IDs, wherein the initial user ID is a unique identifier that characterizes the initial user. Those skilled in the art know that any method of obtaining a user identifier in the prior art falls within the protection scope of the present invention and will not be described in detail herein.

[0023] Specifically, in S100, A is obtained by the following steps: ij :

[0024] S101, obtaining an initial data set, wherein the initial data set includes a plurality of initial data, and the initial data is data obtained from a plurality of initial data terminals.

[0025] Specifically, the initial data terminal is a data terminal for recording information about abnormal physical status of the user.

[0026] Furthermore, the initial data terminal is connected to various related departments in real time through a wireless network to achieve data sharing, such as the initial data terminal such as the pharmacy data terminal and the laboratory data terminal.

[0027] Specifically, for example: initial data such as user ID, gender corresponding to user ID, clinical symptoms corresponding to user ID, etc.

[0028] S102, according to the initial data set, obtain the initial feature information list set D corresponding to the initial user ID list = {D1, ..., D i , ..., D n}, D i ={D i1 , ..., D ij , ..., D im}, where D ij is the jth initial feature information in the initial feature information list corresponding to the i-th initial user ID.

[0029] Specifically, the initial characteristic information is the characteristic information corresponding to the initial user ID under the initial characteristic identifier.

[0030] Furthermore, the initial feature identifier includes words describing the user's physical condition and words corresponding to process nodes in the process of treating the user's body in an abnormal state, such as age, ethnicity, weight, medication status, imaging examination name and results, and other initial feature identifiers.

[0031] Specifically, S102 also includes the following steps:

[0032] S1021, standardize the initial data set to obtain an intermediate data set.

[0033] Specifically, those skilled in the art know that any method for standardization in the prior art falls within the protection scope of the present invention and will not be described in detail herein, for example: methods for standardization using ICD10, test standard dictionaries, etc.

[0034] S1023, taking the user as the primary index, the intermediate data set is associated with D through the primary index.

[0035] Specifically, those skilled in the art know that any method of obtaining associated information by establishing a primary index in the prior art falls within the protection scope of the present invention and will not be described in detail herein.

[0036] Specifically, the initial user is a user name obtained from the intermediate data set.

[0037] S103, when D ij When it is the first type of feature information, ij Discretize to obtain A ij .

[0038] Specifically, the first type of characteristic information is text data. For example, when obtaining the gender, population and medication name corresponding to a user, the obtained characteristic information is all text type data.

[0039] Specifically, those skilled in the art know that any method for discretizing data in the prior art falls within the protection scope of the present invention and will not be described in detail herein. For example, ethnic groups are divided into 1 to 56, where 1 represents the Han nationality and 2 represents the Zhuang nationality.

[0040] S105, when D ij When it is the second type of feature information, ij Normalize to get A ij .

[0041] Specifically, the second type of characteristic information is numerical data. For example, when obtaining the age, weight and electrocardiogram result corresponding to a user, the obtained characteristic information is all numerical data.

[0042] Specifically, those skilled in the art know that any method for data normalization in the prior art falls within the protection scope of the present invention and will not be described in detail herein.

[0043] S107, when D ij When it is an empty set, get A ij =0.

[0044] S200, according to A, obtain an initial feature vector list set C corresponding to the initial user ID list = {C1, ..., C i , ..., C n}, C i ={C i1 , ..., C ij , ..., C im}, C ij is the jth initial feature vector in the initial feature vector list corresponding to the i-th initial user ID, where C ij =(i, j, A ij ).

[0045] S300, according to C, obtain a candidate feature vector list set D corresponding to the initial user ID list = {D1, ..., D i , ..., D n}, D i ={D i1 , ..., D ir , ..., D is(i)}, D i is the candidate feature vector list corresponding to the i-th initial user ID, D ir D i The rth candidate feature vector in D, r = 1...s(i), s(i) is i The number of candidate feature vectors in , where A ij When C ij From C i to form D i .

[0046] S400, according to D, obtain a specified feature vector list set L corresponding to the initial user ID list = {L1, ..., L i , ..., L n}, L i is a list of designated feature vectors corresponding to the i-th initial user ID, wherein L is obtained in S400 by the following steps: i :

[0047] S401, according to A, obtain a candidate feature identifier list G, wherein the candidate feature identifier list includes a plurality of candidate feature identifiers.

[0048] Specifically, in S401, the candidate feature identifiers are obtained through the following steps:

[0049] S4011, obtain an initial feature identifier list B = {B1, ..., B j , ..., B m}, B j is the jth initial feature identifier.

[0050] S4013, according to A, obtain the target priority list F corresponding to B = {F1, ..., F j , ..., F m}, F j For B j The corresponding target priority, where F j Meet the following conditions:

[0051] β j B in A jThe number of initial user IDs whose corresponding initial feature values ​​are not zero.

[0052] Specifically, exp(-β j / n) is the natural base (-β j / n) power.

[0053] S4015, sort the target priorities in F from large to small, intercept the initial feature identifiers corresponding to the first λ target priorities and insert them into G, where λ≤m.

[0054] Specifically, those skilled in the art know that λ can be selected according to actual needs, which falls within the protection scope of the present invention and will not be described in detail here.

[0055] S403, when D ir When the initial feature identifier corresponding to the initial feature value in G is consistent with the candidate feature identifier in G, D ir Insert to L i To obtain L i .

[0056] In the above, according to the initial eigenvalue, the initial feature vector is screened to obtain a list of candidate feature vectors, and based on the priority of the initial feature identifier, the list of candidate feature vectors is screened to obtain a list of specified feature vectors. On the one hand, meaningless data is deleted, and on the other hand, data with little impact on similarity is deleted, which can reduce the amount of computation for subsequent similarity calculations, reduce time costs, and thus improve the efficiency of obtaining similar users.

[0057] S500, according to L, obtain the target similarity list N={N1,...,N i , ..., N n},N i is the target similarity corresponding to the i-th initial user ID.

[0058] Specifically, in S500, N is obtained by the following steps: i :

[0059] S501, according to L i , get the target feature vector M corresponding to the i-th initial user ID i =(M i1 , ..., M ig , ..., M ih(i) ), M ig M i The value of the gth position in the L, g = 1...h(i), h(i) is L i The number of eigenvectors is specified in M, where i The method of obtaining Li The initial eigenvalues ​​of each specified eigenvector in M ​​are inserted into i middle.

[0060] S502, obtain M i The corresponding first key vector Y i =(Y i1 , ..., Y ie , ..., Y iu ), where Y ie Meet the following conditions:

[0061] Y ie =min{η e i1 , ..., η e ig , ..., η e ih(i)}, η e ig =f e (M ig ×M 0 ig ), M 0 ig is M obtained from F ig The target priority of the corresponding initial feature identifier, f e () is the e-th hash function, e=1...u, u is the number of hash functions.

[0062] Specifically, those skilled in the art know that the hash function can be selected according to actual needs, all of which fall within the protection scope of the present invention and will not be described in detail here.

[0063] S503: Acquire a final feature vector corresponding to the target user ID, wherein the method for acquiring the final feature vector is consistent with the method for acquiring the specified feature vector corresponding to the initial user ID.

[0064] Specifically, the target user ID is a unique identifier that represents the target user, wherein the method for obtaining the target user ID is consistent with the method for obtaining the initial user ID.

[0065] Furthermore, the target user is a user whose body is in a specific abnormal state, wherein those skilled in the art know that a user whose body is in any specific abnormal state can be selected according to actual needs, such as abnormal states such as severe fractures.

[0066] S504: Obtain a second key vector X corresponding to the final feature vector, wherein the second key vector X is obtained in the same manner as the first key vector.

[0067] S505, according to Yi and X, get N i , where N i Y i The similarity between and X.

[0068] Specifically, those skilled in the art know that any method for obtaining the similarity between vectors in the prior art falls within the protection scope of the present invention and will not be described in detail herein.

[0069] In the above, the obtained raw data is first preprocessed, and the preprocessed data is screened. Through the distributed computing capability, the preprocessing time of large-scale data is reduced. The hash model is used to obtain the feature vector corresponding to the user, and the similarity judgment is performed based on the obtained feature vector, thereby improving the accuracy of obtaining similar users.

[0070] S600, when N i ≥N 0 , the initial user corresponding to the i-th initial user ID is determined as a similar user of the target user, where N 0 is the preset similarity threshold.

[0071] Specifically, N 0 The value range of N is 0.6 to 0.8, wherein those skilled in the art know that N can be adjusted according to actual needs. 0 The selection of all falls within the protection scope of the present invention and will not be repeated here.

[0072] The present embodiment provides a data processing system for determining similar users, including a processor and a memory storing a computer program. When the computer program is executed by the processor, the following steps are implemented: obtaining an initial feature value list set corresponding to an initial user ID list, obtaining an initial feature vector list set corresponding to the initial user ID list based on the initial feature value list set, obtaining a candidate feature vector list set corresponding to the initial user ID list based on the initial feature vector list set, obtaining a designated feature vector list set corresponding to the initial user ID list based on the candidate feature vector list set, obtaining a target similarity list corresponding to the initial user ID list based on the designated feature vector list set, and determining similar users of a target user from the initial user ID list based on the target similarity list. In the process of obtaining similar users, the present embodiment deletes meaningless data and considers deleting data with little impact on similarity, which can reduce the amount of calculation for subsequent similarity calculation and reduce time cost, thereby improving the efficiency of obtaining similar users, preprocessing the obtained original data, obtaining feature vectors corresponding to users using a hash model, and performing similarity judgment based on the obtained feature vectors, thereby improving the accuracy of obtaining similar users.

[0073] Although some specific embodiments of the present invention have been described in detail by way of example, it will be appreciated by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the present invention. It will also be appreciated by those skilled in the art that various modifications may be made to the embodiments without departing from the scope and spirit of the present invention. The scope of the present invention is defined by the appended claims.

Claims

1. A data processing system for determining similar users, characterized in that: The system comprises: a processor and a memory storing a computer program, and when the computer program is executed by the processor, the following steps are implemented: S100, obtaining an initial feature value list set A corresponding to the initial user ID list = {A1, ..., A i , ..., A n }, A i ={A i1 , ..., A ij , ..., A im }, where A ij is the jth initial eigenvalue in the initial eigenvalue list corresponding to the i-th initial user ID, i=1...n, n is the number of initial user IDs, j=1...m, m is the number of initial eigenvalues ​​in the initial eigenvalue list; S200, according to A, obtain an initial feature vector list set C corresponding to the initial user ID list = {C1, ..., C i , ..., C n }, C i ={C i1 , ..., C ij , ..., C im }, C ij is the jth initial feature vector in the initial feature vector list corresponding to the i-th initial user ID, where C ij =(i, j, A ij ); S300, according to C, obtain a candidate feature vector list set D corresponding to the initial user ID list = {D1, ..., D i , ..., D n }, D i ={D i1 , ..., D ir , ..., D is(i) }, D i is the candidate feature vector list corresponding to the i-th initial user ID, D ir D i The rth candidate feature vector in D, r = 1...s(i), s(i) is i The number of candidate feature vectors in , where A ij When C ij From C i to form D i ; S400, according to D, obtain a specified feature vector list set L corresponding to the initial user ID list = {L1, ..., L i , ..., L n }, L i is a list of designated feature vectors corresponding to the i-th initial user ID, wherein L is obtained in S400 by the following steps: i : S401, according to A, obtaining a candidate feature identifier list G, wherein the candidate feature identifier list includes a plurality of candidate feature identifiers; S403, when D ir When the initial feature identifier corresponding to the initial feature value in G is consistent with the candidate feature identifier in G, D ir Insert to L i To obtain L i ; S500, according to L, obtain the target similarity list N={N1,...,N i , ..., N n },N i is the target similarity corresponding to the i-th initial user ID; S600, when N i ≥N 0 , the initial user corresponding to the i-th initial user ID is determined as a similar user of the target user, where N 0 is the preset similarity threshold.

2. The data processing system for determining similar users according to claim 1, characterized in that: In S100, A is obtained by the following steps: ij : S101, obtaining an initial data set, wherein the initial data set includes a plurality of initial data, and the initial data is data obtained from a plurality of initial data terminals; S102, according to the initial data set, obtain the initial feature information list set D corresponding to the initial user ID list = {D1, ..., D i , ..., D n }, D i ={D i1 , ..., D ij , ..., D im }, where D ij is the jth initial feature information in the initial feature information list corresponding to the i-th initial user ID; S103, when D ij When it is the first type of feature information, ij Discretize to obtain A ij ; S105, when D ij When it is the second type of feature information, ij Normalize to get A ij ; S107, when D ij When it is an empty set, get A ij =0.

3. The data processing system for determining similar users according to claim 2, characterized in that: The initial data terminal is a data terminal for recording abnormal physical status information of the user.

4. The data processing system for determining similar users according to claim 2, characterized in that: The initial characteristic information is the characteristic information corresponding to the initial user ID under the initial characteristic identifier.

5. The data processing system for determining similar users according to claim 2, characterized in that: The first type of characteristic information is text data.

6. The data processing system for determining similar users according to claim 2, characterized in that: The second type of characteristic information is numerical data.

7. The data processing system for determining similar users according to claim 1, characterized in that: In S401, the candidate feature identifiers are obtained through the following steps: S4011, obtain an initial feature identifier list B = {B1, ..., B j , ..., B m }, B j is the jth initial feature identifier; S4013, according to A, obtain the target priority list F corresponding to B = {F1, ..., F j , ..., F m }, F j For B j The corresponding target priority, where F j Meet the following conditions: β j B in A j The number of initial user IDs whose corresponding initial eigenvalues ​​are not zero; S4015, sort the target priorities in F from large to small, intercept the initial feature identifiers corresponding to the first λ target priorities and insert them into G, where λ≤m.

8. The data processing system for determining similar users according to claim 1, characterized in that: In S500, N is obtained by the following steps: i : S501, according to L i , get the target feature vector M corresponding to the i-th initial user ID i =(M i1 , ..., M ig , ..., M ih(i) ), M ig M i The value of the gth position in the L, g = 1...h(i), h(i) is L i The number of eigenvectors is specified in M, where i The way to obtain L i The initial eigenvalues ​​of each specified eigenvector in M ​​are inserted into i middle; S502, obtain M i The corresponding first key vector Y i =(Y i1 , ..., Y ie , ..., Y iu ), where Y ie Meet the following conditions: Y ie =min{η e i1 , ..., η e ig , ..., η e ih(i) }, η e ig =f e (M ig ×M 0 ig ), M 0 ig is M obtained from F ig The target priority of the corresponding initial feature identifier, f e () is the e-th hash function, e=1...u, u is the number of hash functions; S503, obtaining a final feature vector corresponding to the target user ID, wherein the method for obtaining the final feature vector is consistent with the method for obtaining the specified feature vector corresponding to the initial user ID; S504, obtaining a second key vector X corresponding to the final feature vector, wherein the second key vector X is obtained in the same manner as the first key vector; S505, according to Y i and X, get N i , where N i Y i The similarity between and X.

9. The data processing system for determining similar users according to claim 1, characterized in that: N 0 The value range is 0.6~0.8.

Images